EasyAwscr::S3::Client

From time to time, we should recreated the connection pool, so it will reload the SSL context (see https://github.com/crystal-lang/crystal/issues/15419).

This is a hard limit enforced by AWS for multipart uploads: each part must be at least 5 MB.

Converts an existing client from "awscr-s3" to use the "easy-awscr" client interface.

Aborts a multi part upload. Returns true if the abort was a success, false otherwise.

resp = client.abort_multipart_upload("bucket1", "obj", "123")
p resp # => true

Batch deletes a list of object keys in a single request.

resp = client.batch_delete("bucket1", ["obj", "obj2"])
p resp.success? # => true

Closes this client. If used again, a new connection will be opened.

Complete a multipart upload

resp = client.complete_multipart_upload("bucket1", "obj", "123", parts)
p resp.key # => obj

Copy an object from source to destination in a bucket.

client.copy_object("bucket1", "source_object", "destination_object")

Delete a bucket, note: it must be empty

resp = client.delete_bucket("test")
p resp # => true

Delete an object from a bucket, returns true if successful, false otherwise.

resp = client.delete_object("bucket1", "obj")
p resp # => true

Get the contents of an object in a bucket

resp = client.get_object("bucket1", "obj")
p resp.body # => "MY DATA"

Get the contents of an object in a bucket as an IO object

client.get_object("bucket1", "obj") do |resp|
  IO.copy(resp.body_io, STDOUT) # => "MY DATA"
end

Get information about a bucket, useful for determining if a bucket exists. Raises a Http::ServerError if the bucket does not exist.

resp = client.head_bucket("bucket1")
p resp # => true

Get the metadata of an object in a bucket

resp = client.head_object("bucket1", "obj")
p resp.size          # => 123
p resp.status        # => HTTP::Status::OK
p resp.last_modified # => "Wed, 19 Jun 2019 11:55:33 GMT"
p resp.etag          # => ""
p resp.meta          # => {"my_tag" => "my_value"}

List s3 buckets

resp = client.list_buckets
p resp.buckets.map(&.name) # => ["bucket1", "bucket2"]

List all the items in a bucket

resp = client.list_objects("bucket1", prefix: "test")
p resp.map(&.key) # => ["obj"]

Create a bucket, optionally place it in a region.

resp = client.create_bucket("test")
p resp # => true

Add an object to a bucket.

resp = client.put_object("bucket1", "obj", "MY DATA")
p resp.key # => "obj"

Start a multipart upload

resp = client.start_multipart_upload("bucket1", "obj")
p resp.upload_id # => someid

Provides IO that can be used to stream directly into an S3 file. In contrast to upload_file, the size of the data does not have to be known before. It will use start_multipart_upload, upload_part, and complete_multipart_upload internally.

Example: creates a file on S3

client.stream_to_s3("bucket1", "obj") do |io|
  io << ...
end

Intuitively, it is like writing to a local file, matching this pattern:

File.open("/tmp/bucket1/obj", "w") do |io|
  io << ...
end

If you need more control, you can also get direct access to the IO object:

io = client.stream_to_s3("bucket1", "obj", auto_close: false) { |io| io }
io << ...
io.close

Upload a file to a bucket. Returns true if successful; otherwise an Http::ServerError is thrown.

File.open("/path/some/big/file.txt") do |io|
  success = client.upload_file("bucket1", "obj", io)
  p success => true
end

It uses Awscr::S3::FileUploader internally:

• If the file is 5MB or lower, it will be uploaded in a single request; but if the file is greater than 5MB, it will be uploaded in parts.
• If with_content_type is true, the uploader will automatically add a content type header

Upload a part, for use in multipart uploading

resp = client.upload_part("bucket1", "obj", "someid", 123, "MY DATA")
p resp.upload_id # => someid